Joshua Shay Kricheli NeuroSymbolic AI
v2 v1

Release · v2.9.3

Rebuild seasonal background as a flow-field weather…

Rebuild seasonal background as a flow-field weather simulation

chore v2.9.3 June 21, 2026 Claude 3c5fdcb

Details

Replace the simple per-season particle effect with a real, multi-layer
physics simulation:
- Wind is a time-evolving curl-noise flow field (divergence-free, so it
  swirls like real air) that drifts between laminar and turbulent
  regimes with periodic gusts. Inline Perlin noise; no dependencies.
- Particles are advected through the field with gravity, aerodynamic
  drag toward the local wind, buoyancy and flutter, so motion is
  emergent rather than scripted. A coarse grid caches the field and
  particles bilinearly sample it for performance.
- Each season composes ~10 mild layers: parallax particle bands (depth
  via per-layer wind response/size/alpha), occasional precipitation or
  flurry passes (winter sleet/rain mix, spring/fall gust flurries,
  summer dandelion seeds), plus atmosphere (washes, glows, god-rays,
  ground mist, vignettes).
- Winter reads as wind-blown snow that occasionally mixes to sleet/rain;
  summer is a sun with rays + warm wash; spring/fall flutter with
  petals/leaves of varied species and sizes.
Still theme-aware, pointer-events:none, density scales with viewport
and device, pauses when hidden, and disabled under reduced-motion.

Files changed (1)

frontend/client/ts/seasonal-background.ts +589 −327